def test_clf_with_nan(self, input_data, expected_value):
params = input_data.copy()
vals = {}
vals["y_pred"] = params.pop("y_pred")
vals["y"] = params.pop("y")
metric = ConfusionMatrixMetric(**params)
result = metric(**vals)
np.testing.assert_allclose(result, expected_value, atol=1e-4, rtol=1e-4)
result, _ = metric.aggregate(reduction="mean_channel")[0]
expected_value, _ = do_metric_reduction(expected_value, "mean_channel")
expected_value = compute_confusion_matrix_metric("tpr", expected_value)
np.testing.assert_allclose(result, expected_value, atol=1e-4, rtol=1e-4)
def compute(self):
"""
Raises:
NotComputableError: When ``compute`` is called before an ``update`` occurs.
"""
if self.compute_sample is True:
if self._num_examples == 0:
raise NotComputableError(
"ConfusionMatrix metric must have at least one example before it can be computed."
)
return self._sum / self._num_examples
confusion_matrix = torch.tensor(
[self._total_tp, self._total_fp, self._total_tn, self._total_fn])
return compute_confusion_matrix_metric(self.metric_name,
confusion_matrix)
def _reduce(self, scores) -> Any:
confusion_matrix, _ = do_metric_reduction(scores, MetricReduction.MEAN)
return compute_confusion_matrix_metric(self.metric_name, confusion_matrix)